An IoT and Random Forest-DT-Based Early Enhanced Forest Fire Detection in Large and Small Targets

The world is confronted with more complex and severe environmental concerns than in the past. Extreme weather events have become more common and destructive in the last several decades, endangering forests and urban forests around the globe. Massive forest fires are among the most catastrophic natural catastrophes that affect Earth's climate and the diversity of life on it. Thus, in order to minimize harmful impacts on ecosystems and people, it is vital to have well-planned and coordinated plans for early warning, prevention, and reaction. Image preprocessing, feature extraction, and training the model are the initial steps in the suggested methodology. Histogram matching and picture smoothing are part of the image preparation procedure. The gray histogram is a way to see how many pixels fall into each grayscale level, and image smoothing is a way to enhance a large region of a picture using image processing. The GLCM and LBP histograms are employed in the feature extraction process. Training RF-Multivariate ARS models requires feature extraction as a prerequisite. The system improved accuracy to 97.29 percent after applying the approach.